Batteries are increasingly central to transportation and the electric grid, yet many of the processes that determine their lifetime, safety, and performance remain hidden from conventional measurements. In this talk, I will discuss two complementary approaches for gaining access to this hidden information. First, cryogenic electron microscopy enables direct visualization of battery interfaces and degradation processes at the nanoscale. Second, machine-learning-based virtual sensing can infer internal electrode states from measurements available outside the battery, creating a “virtual reference electrode” that can improve state estimation, detect lithium plating, and enable smarter charging. Together, these approaches point toward a future in which batteries are not only better understood from the inside, but also become increasingly intelligent in how they are monitored and controlled.